One transformation, two moves: modernize the platform, infuse AI across everything we build.

A functional but rigid legacy estate priced every change in months.

The transformation began against a fixed-cost mainframe estate with deep vendor dependency and a shrinking specialist talent pool. The platform worked — but every product change moved at legacy speed, and cost scaled linearly with membership.

Baseline from the January 2026 strategic business case and the April 10, 2026 mainframe migration assessment kickoff.

From legacy mainframe to an AI-native, cloud-native platform — one platform at a time.

The strategy is a strangler pattern: build new cloud-native platforms off the mainframe and peel dependency away platform by platform, so the legacy estate shrinks until it can be switched off. We run it two ways at once — building new products with agentic development on our AI SDLC, and using AI to modernize the legacy code itself.

Where we are now: the new platforms run cloud-native on their own PostgreSQL, but still depend on the mainframe database via CDC sync and backend batch processing for payments and communications. The next move is event-driven, real-time capability that retires the batch dependency and lets the legacy estate die on the vine.

Three 2025 pilots proved the model — all completed, and each now carries into the platform program.

Before scaling, the team ran three pilots in 2025, each proving a different part of the approach: build greenfield, modernize legacy code with AI, and ship new regulated capability. All three completed successfully and graduated into the production platform program.

Three pilots completed in 2025 (per the AI Factory strategy deck); pilot metrics as reported at completion. They establish the capability the 2026 platform program builds on.

$0.60 per member per month — a 52% run-rate reduction on an evidence-gated path.

Savings are counted when workloads are verifiably retired — not when code is migrated. That discipline keeps the case credible.

Source basis: WHPS AI Transformation strategic business case, January 2026.

A phased path from foundation to mainframe exit — with wins at every stage.

The program runs two tracks at once: platform modernization and AI transformation. Each phase is expected to land usable, evidence-backed capability — so value compounds along the way rather than arriving only at the end.

Wins along the way — not a single payoff at the end of a multi-year arc.

The WHPS AI SDLC Factory: Agile discipline adapted for agentic delivery.

Delivery speed comes from a documented, model-agnostic operating method — not from any single tool. Agents execute bounded work inside scoped workspaces; humans own the gates; every release ships with a typed evidence packet. The method itself advances with each initiative it delivers.

Pilot results across the 2025 builds: 180% faster development and 65% fewer defects versus traditional delivery, as reported at pilot completion.

Security is a release condition, not a late-stage advisory step. See the full AI SDLC control plane and diagrams →

Every initiative leaves a reusable AI-native capability behind.

The initiatives are not just products. Each one proves and hardens a durable capability that the next build reuses — so the program moves toward an AI-native operating model with every delivery, not only at the finish line.

Capabilities describe the target AI-native operating model the program is building toward as initiatives complete.

One governed register: sixteen AI initiatives, grouped by capability family.

This is the whole portfolio at a glance — from the platform builds to the Contact Center AI family, reconciliation, and the operational AI tools. The following chapters walk each family in turn.

Statuses per the July 1, 2026 portfolio register. Dot colour: green live · blue active · red at risk · amber decision · grey queued. Open the full register page →

Speed that leaves an audit trail.

Every requirement row traces from CMS source through backlog, delivery gate, test kit, and evidence reference. The evidence system was developed with the platform, so audit readiness is part of delivery — and no external claim outruns the evidence.

Traceability coverage and auditor evidence completion are tracked as separate statuses — current FIT evidence capture is in progress, and external claims are timed to completed audit and certification evidence.

The proven delivery engine now points at the mainframe exit.

GroupLink's coexistence bridge de-risks the DB2 migration waves. The MarketLink pathway proves regulated delivery at AI SDLC speed. The AWS modernization proposal — preview received June 24, 2026, with the formal proposal as the next decision input — frames the sequencing decision now on the table. The same platform is designed to be future-ready for new markets: ICHRA, Medicare, and Medicaid.

Decision velocity now determines scale. See the vendor decision room →